Epifun
Toulouse, le 7 Décembre 2018

The Omnigenic Theory for the Genetic Architecture of Complex Traits

Does this theory hold in other biological kingdoms?

  • Can we test the omnigenic theory with empirical data in Poplar?

  • Approach:

    • Build gene networks from transcriptomic data to define core genes

    • Complex traits prediction

    • Quantitative and Population genetics of gene expression with respect to their position in the network

Material, Experimental Design & Data

  • 240 genotypes from 11 P. nigra natural populations
  • Two common garden experiments (RCBD, 6 blocks):
    • ORL (Fr)
    • SAV (It)
  • Contrasted growth conditions:

RNAseq data collection and processing

  • ORL design, June 2015 (4th year of growth)
  • Cambium and Xylem on 480 trees (2 biol. reps. in 2 blocks) \(\rightarrow\) 960 RNA samples
  • Equimolar pooling of RNA per tree
  • Sequencing @ POPS transcriptomic platform
    • Illumina HiSeq 2000, Single-Read
    • 10 samples / lane \(\rightarrow\) \(\sim 20\) million reads / sample
  • Mapping on P. trichocarpa v3.0 41,335 primary transcripts \(\rightarrow\) read counts
  • Data cleaning pipeline:
    • Filter out transcripts with 0 count in any sample (n = 1,653)
    • 39,682 transcripts : Normalization (TMM)
      Variance stabilization (\(log_2(n+1)\))

Building a gene network

  • R package WGCNA (Langfelder & Horvath, 2008)

Defining Gene Subsets

  • Module membership (\(KME\)) provides a measure of connectivity within the network
  • Three \(KME\) based gene subsets
    (\(n = 3,968\) - \(10 \%\))
    • Core
    • Peripheral
    • Random

  • One ‘statistically important’ subset (\(n = 635\)) Boruta (Kursa & Rudnicki 2010)

Which Genes Matter to Predict Complex Traits?

  • R package h2o (Kraljevic, 2018)
  • Two constrasted prediction models

Which Genes Matter to Predict Complex Traits?

Core and Peripheral Gene Subsets Differ Depending on the Prediction Model Used

–>

Core and Peripheral Gene Subsets Differ According to their Genetic Features

Summary

Acknowledgements

  • V Jorge, O Rogier (INRA BioForA, Orléans)
  • V Brunaud, ML Martin Magniette, C Paysant-Leroux, L Soubigou-Taconnat (POPS transcriptomic platform @ IPS2, Saclay)
  • ANR - SYBIOPOP project (ANR-13-JSV6-0001)